Machine-Learning Phishing Detection Model Used in the E-Banking Environment

The exponential expansion of Internet usage has given rise to a significant upsurge in cyberattacks, which have caused damage to brand reputation, privacy and financial information, and identities. Phishing, an enduring cyber threat, has emerged as a substantial concern because it can cause considerable financial detriment to economies and erode users’ confidence in e-commerce and online banking. The purpose of this research is to identify phishing websites through the development of a Phishing URL Detection Model (PUDM) utilising machine learning techniques. The model utilises machine-learning algorithms and a standard UCI machine-learning library dataset. The performance of the proposed methods surpassed that of several machine learning algorithms when compared to related work for this study. A feature importance analysis was performed to ascertain the features that would be employed to distinguish phishing URLs from authentic ones. The study determined that the Google Index feature had the greatest impact on determining the validity of website URLs. In contrast, the XGBoost classifier demonstrated the highest performance, attaining an F1 score of 92.72%. The results of this study have the potential to significantly enhance the security measures in place for organisations, clients, and website proprietors.